(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "COMPARISON OF THE RELIABILITY AND VALIDITY OF VISUAL ACUITY TEST TARGETS"

SCHOQL OF AVIATION MEDICINE 

tJU^S. NAVAL AIR STATION 
PENSACOLA FLORIDA 




A C0MPARISC9J OF THE RELIABILITY AND VALIDITT OF 
VISUAL ACUITY TEST TARGETS 



Reaoarch Project X-676 {AT-357-p} Report No, 1 



3 April 1946 



IIAVAL 3GH00L OF /lVIATIOK kEDIGIKE 
U. 3. NAVAL AIR TRAINIl'G BA3E3 
PiiliGACOL/., J1,0RIDA 



liE3EARCH REPORT 
amiTTED - 3 APKIL 1S46 



PROJECT HO, X-676 (Av- 3£7~p) 



REPORT JJO. Om 



TITLE: 



REPOKT BY: 



^9^'}^M^PPI'^L RiaiABiiiTY j^m validity of 

VIoUAL AGtilTY TEST T/iliOETS. 





Lt. iiichard Trwiibull , 11(3} , U3KR 
Lt. {jg) Oscar PQckstroB:., Jr 11 ( sl'jf IS^'R 





GAF1V\II%' LOUIS IVJiff{30K, (MG) , ISN 
KJ5DICAL OmCER W CliARGE , 



The validity of fiv« targets for testing visual acuity was de- 
termined through the correlation of test results of each target with 
the Grow chart aoores. In addition, the teat-reteat reliability of 
all six tests was studied. The targets under consideration were: 
The 3ight Screonor, a Nets London letter target, two Randolph Field 
letter targets , and an adaptation of the Ortho-Rater checkerboard 
type target. One hundred subjects were employed. Acuity scores, ob- 
tained In Jnellen equivalents, were translated into log-units to fac- 
ilitate statistical analysis. Additional systems were essayed for 
scorine each of the liandolph Field testa, Statistical analyaes con- 
sisted in {1) deriving teet-retest reliability coefficients and co- 
efficients of Intercorrelation among the test scores by means of pro- 
duct-woment correlation, (2) determining the significance of the dif- 
ferences between test and reteat mean scores, {3) determining the alg- 
nifieonce of the differences betvreen mean scores on the various tests 
by the _t tests, and (4) obtaining the conventional measures of disper- 
sion and statistical reliability. Results of the statistical analyses 
and clinical observations of the exaBiiners relating to each test are 
presented, - 

CQHGUJSIOKa 

1. With the exception of the Ortho-Rater type target and the Sight 
Screencr, which have somewhat lower overall reliabilities, there is 
little practical difference in overall test-retest reliabilities among 
the various targets. 

2. There la no important difference aiaong the targets in correlation 
with the Grow chart except for the Ortho-ltoter and Sight Screener 



targets, whose correlations with the Grow chart are conaiatently 
lower. 

3. Subjects show slight but statistically reliable gain on retast 
with all targets. 

4. Some targets, the Randolph Field 1-unit in particular, consist- 
ently tend to give higher mean scores than others. < 

5. iiesults shown by the Randolph Field 1-unlt decrement target in- 
dicate that finer quantitative testing of visual acuity is possible 
where job analysis requires it. This target also has an advantage 
in ternss of ease smd speed of administration. 



IKTRODTOTION 

This project was eatabliahed to stMdy the reliability fin3 valid- 
ity of five targets for testing visual acuity. Reliability was deter- 
in ined by a teat-retest coiaparlaon while validity ma doterEitied using 
the Grow chart as the criterion. Of the five targets, four ware nount- 
ed upon glaos panels for ja^eaentatioij in a cabinet* developed for trans- 
ill wiiinati oh Of test tereeta. These targets include three of the latter 
type, a target developed and validated at the Hew London aubrffirine base 
and two targata developed and validated at Handolph Field by the^ AAF. 
The fourth target was an adaptation of the Ortho-Hater type test tar- 
get, enployine checkerboard pattern diacrlKlnation ea a ireaaure of acuity. 
The reniaining target, presented in the Sight Screenert* waa the "distance 
acuity'* portion of the slide supplied with that Instrun^nt. ^S'ith the 
exception of the latter the targets were developed for use froni a 30* 
distance and adapted to •walk-up" procedure similar to that erftployed by 
the Kavy with the Grow chart. Every effort waa made fvirther to equate 
these targets -with the Grovj in terria of illui-'-ination, procedure, et cetera. 

mocmRm ■ 

EKperiJKen tal Deaign i 

In order to elijf.lnato any undue influence of one test "upon another 
developing fron its place within the series, as well as differential ef- 
fects of fatigue, the teats vjere presented in rotation. Considering the 
three trans-ill urdna tad letter targets aa one type essentially different 
froPi the O-row Chart, Jlght Screenor, and the Ortho-Rater checkerboard 

tarisets , it vjaa poar.ible to desi gn a rot a ti or; of projjentatlpn wjj^^ four 

* A product of the Bausch X Lornb Optical ^onipany 
*♦ A product of the American Optical Companj' 



units. The four units resulted in 24 orders of presentation, with each unit 
appearing in e&ch serial position with equal frequency. In addition the or- 
der of preaentati-in within the trans-illuminated letter group also was ro- 
tated. The order of presentation was the sai!:e for any one subject on both 
test end rate at. iietests ware scheduled after a siininiuiri interval of 46 
hours; the group average interval was 78 hours. All wrong responses were 
recorded upon anavjer sheets so constructed tliat values could be datenninoil 
easily. 

dubjects: There vnere 100 subjects used in this study. An effort wa3 made 

to obtain n ropresnntatiira group of acuity values,- as indlteatod by ^corea 
on the Grow chart. It la obvious that a service group is preselected and 
not truly repTesantative of the population as a whole. This factor notwith- 
standing, Table 1 shavjs to \>tl)&.t extent this group does correspor.d with trat 
of previous studies (1,2) which Indicate that ^5fi, -^bfo of the population have 
leas than 20/20 vision without glasses. The effort was it-ade to obtain such 
a group, althou^^h this ivas not a normative study, .because 'it was. believed 
that tha ef f icienc;. ;!nd reliability of the testes should be aeterE-,ined by it!- 
cludLng reproaantative levels of ..rjubnornal acuity as ivell as iO/^O acuity - ' 
end above, Isini; imlei subjects only, the population wa.'3 taken fror;: 

1. Volunteers fro;:, the officer .-^nd unlisted personnel attached to 
the J'.aiii Dispensary at Fensacola. ■ 

ii.. Jtudents fxom a class of Hospital Corpanien under training as Avi- 
ation fcediaine Technicians.' ' " ' , 

3. litolisted personnel appearix^g for refractions at the ^ye Clinic prior 
to adii;ii:i3trati-jn of hojnati-opine . 
The age range represented by this group was frois: 18.5 to 45 years .'dth a me- 



dlan of -sol- years, Kode of 19. C years and r. eun (^^e of years B tionths. 
Kqui^atxt jiiid teJtiKj; p. e t hod ; The equJ piKent included a ihodified form of 
the Kavy Grow <shart, the 3ight Sereeper and four trcipg-illujfiiralied test 
t&j'get3*u$ diacuaaed above and sham in Fig, 1-4. 

Duritte original plannlPt; of proctidure, cej*tain criteria vi'ere es- , 
tiiblished foj* teiiriiiatipg tqatin^ on any teirget. This procedure' --.as dl3- 
cofitinuad, however, wliein it was found thst iiubjeets v.-ho would have beon 
ytopp^d at a oeftaic level- on thia bgsia -Aare able to cotiticue and read 
3liU3(3queAt lines duccoaaf uily , (3eo diacviaair.n of teats). For this rea- 
.son tli©' teatir.g oii Jinj'- target v/aa allowed to ^ontinvie until the aybject- 
'.vjia itjr;.-able, to ;roc**0d. Re wa-i er^cQurij;;;ed to do the.bost !.te could, the 
9X£iii-J,Rer aLiatiiiing' the reriponaibili ty for enooixrefjir.e thoac v-ho had a tend-' 
enoy to "givf yp'' eaasiiy and al90 for tarit.iR«iting taata, . During rpnooular 
tests, the ;>ubj^td rterc giver, a statidai-d !; oiiocular eye Jjcreen of 6lun:iiii.an 
L'jhieh bridged tho rose acd could be, ri-^ersed to accorjr.odate tlie ey^? belr^: 
teatad^ (Fig. No aubjocta v^ere teytad -Ahile .searing a refi-activa cor- 
recti or;. ■ 

The levels of ■ acuity ' tatitod by thiJ b«ittery of tart;etf3 a^a prsssntad' 
ir;' Tabltci S. ■ It will be i.oted that ..the KtfK lor don tost varies sojtiawh&t 
froji.- the Vftluea in the loft hand colur-in. The levela irsdicat^d as teatfid 
by tha. Ivsvy Grow chart are thoaa established by the ''imlk-up" positloi.a 
Ghojoi; to correjijond .'.1th' lovel.:5 -of the other targets. The rSght tema 
eolurji iij.£i ioc^trlthrilc trar.tilatior of acuity vi^lues coiiputcd to facj.li- 
tote dtatisticcl ajialyaia. Log scoi-e ^ 100 x log^^(10 x VA} "whei'e ?A 3rsai:3 
Viai!;;il Angle. 

Ay ^ .QtOVi G^UiRT (Fig. The st arAp M ]:ev;/_ Gr pw_ c h jir t ^ c prjalata of ^^e ven 
* iTodiiced by Bjiusch &. Lorc.b Optical Goriij^Jiiy 



lines of 11 charactei's to be read at 20' distance for a 20/20 Jhellen 
rating. Two lines with 11 characters are available for testing 20/15 
vision at 20' and one lire of 11 cJiaractera for teating 20/10 vision at 
<J0', the acuity values of these three lines being established, by a dec- 
rement in size of the test ohapactera. Two charts were prepared with 
alternate lines covered by a grill of .hits cardboard of the sm-e reflec- 
tive value as the background of the teat card. This modification la in 
keeping with recomiKendationa for presenting charta of this nature. 

The teat ivaa given in accordance with Tavy jarocedure, vision firat 
being tested at 20' with subsetiuent walking forward to predetermined 
poaitiona La:til a line could be read .vith two errors or less. To min- 
isiize the ir.eiiiory factor, i-;.:re tiian one lire was used or lines were read 
backward. For low levels of acuity the "walk-up" system presented pos- 
itions equivalent to Snellen 20/22, 20/24, 20/36, 20/28, 20/SO, 20/32, 
20/34, 20/36, 20/38, 20/40, 20/50 and 20/100. 

The illunlnation falling upon the target viiaa 34.0 apparent foot can- 
dles, coming from a 200 watt bulb located 1 liietor in front of and 1 meter 
above the center line of the target. This illuBination and the above 
.Snellen equivalents were used to equate the illurrdnation and levels of 
acuity dlscriminatior v;ith those of the other targets employed in the study, 
Jcoring of the Grow target followed the Navy practice of oreditlrg the line 
of highest acuity read with two errors or less. 

E. aiGIfP 3CREENSR : This standard apparatus (Fig. 1) employs polarl- . 
zation contraat as an occluding method. It measures both monocular and bi- 
nocular visual efficiency vjhile both eyes remain open. Only the distance 
acuity section of this teat was used, the values tested being shown in Table 2. 

-4- 



The subject is seated before the inatruroaiit with hia head ateadied 
by a head i-eot. The test line, located on a slide, is iioaitiouod by ireans 
of a knob which is operated by the exaalnor, 

scoring of the iiight Soreener teat waa confined to one system, that 
of crediting the highest value at which one or no error was made in four 
responses. The values represented by fewer than four character!? (1 eEcli 
for hO/ZOO and 20/100; 2 for 20/70) were credited only when no error was 
rade. 

0. T m'i J-ILLTO' Jlj ATHID TARGETS ; The four trans-llluj.'.inated type ter- 
gets include the Hevi London test (Fig. 3), a Handolph Field teat which 
©jr ploys a decreiient of one urjit par character (Fig. 4), a Randolph Field 
test with a S-unit decrement for groups of three characters (Pig. 4) , and 
the checkerboard test of acuity (Fig. 3), 

In a trans-illuEinated type of target, diffuse light paaaea through 
glass plates upon which the test objects or characters are mounted. In 
contrast, the atiellen and Grow type are illnjcinated by reflection. Trana- 
illuinination has been employed previously In various apparatus such aa tfae 
Ortho-Rater, dight Jcreener, arid early etteuipta at letter chart llluiiinatlon. 
The test targets used in this study were inade on photographic film and 
placed between two plates v;hleh were than sealed as a unit. 

The plates were presented in cabiiiets ..hich were developed for thia 
purpoae. These cabinets (Fig. 3) are approxiKately 36 ' x 27" x 8". They 
ere divided into halves with seirdcylindrical backgrounds for noar-uiilform 
reflection throughout froc the 30 wutt white fluorescent bulb which Is placed 
in the front center of the cabinet. The resultant uniforcilty of illui/ilnatior. 
is shown in Fig. 3, where one half of a cabinet is open. Sach half of the 
cabinet is further divided into two openirga of 10|"x 13|': Guides perrrdt in- 

-5- 



sertion of targets such as the titove in a centerecl position. Grse eabiret 

VV&3 jiifjidified (Fig. 4) for this at'jdy in order to present but one target 

while Biairtainirg conatant surroui-ici , by the raCkUtion of four gate panels 

whicti could occlude any cor.bii ation of targets; . 

Of the foui- isrsria-illiurdrated tests preaenttd ir thefje cabineta, 

two consisted of two plates while tlie other two ivcre corifiried to. a 3ln;v.l8 

piste euch. Thus, there were two opeiiiif;3 in ore cabinet v.hich were oc- ■ 

eluded durlrg this study. The positiar.s i'n which the targets were pi'ced 

(Fig. 1) were established after a stvdy cf the apparert foot cundle il- 

lurdnations of the four openings of each cabinet as well as ths trans - 

mission praperties of the indiTldual target plate-ia. The afc values of 

the plates in the position u;-3c.d ars-? prea-snted belov;, each figure repre- 

-Ssnting an average of £0 readirga: 

Plate afc_ 

Top Checkerboard 39.0 
Lower Checkei-board 40.0 

I'dandolph 3-Unit 40.0 
Randolph 1-Unit 3?. 5 

Top Itew London , ' 41.0 ' • 

Lower New London " 40.0 

TLese targets were r€;ad I'roi the £0' distanc?3, with 16', iO' aci, s 

in a feiv cases, fj' positions used for tjesting loi'; !;:cuitis3. Th« last 

position Eientioned was used prir.ar'U:/' to eljcinate the raeir.ory f&ctcr. 

Less than DOfh of the >jubject& iH-.iiirsa '-ralk-.ip'' i.-oxe than oxis tar- 

jist. The '» walk-up" positions vnere uasd of.iy -.shea, necessary, the -jubi^ict 

vet'cxLirg to the 20* position for suLasque'it tsst^, 

1. MI i£S2PA l^i^T: (Fie.. 5) 

The values of acuity for the Ijnea of tihip test are sho'sc in Tab^c 

The subject, standing; at the £G» po,j.lt;lon, o,-!-. luded left eye i^sii;:; 



tho shield. The test targets (two platea) ware exposed and he wq3 , 
asked to read the letters. 

All lines with 8 ehapactera and the 20/33.3 line cf 7 characters 
-mvB cjredited when S errors or loss were modo. The last line ffiseting 
this r«Quireni(jnt was crediied although previoua lines might have beon 
vm.& -J/irA\. wore than 2 ©rrors. Other lines ware credited with 1 error 
or jao «rror, with the exception of the 20/200 line of KG whieh never 
was ersdited beeauss it viaa memorized so readily. 

^" i^-^jQOIPH niSLD 1-Unit ; ' • , ■ 

This tai'gat, shown in Fig. 4, haa sinele .characters' rspreseiitijig 

ijp.it dscr?ji!;ent levels from ao/40 tq 20/5, The subject was, tos'ted in the ' 

uaual Fianner at 20' and was icoved forward to ''walk-up" positions onlf 

under circuisstancea previously discussed, 

Tvio scoring ayatema were eii;;ployed for this t^st ; 

1, The aoulty value of the last character correctly identified. This, 

eatsiblislied tho irsxiisiurs acuity, 

£, Thi? sicuity value of the character preceding; the firat failure on 
two ch.aractei'3 in aueceaaion.' The, relative itierits of the two systeas^ 
ir. terxi'ts of reliability and validity, are showi below. ' ' 
3* l^^-^.DOX'H^ ^m.D 5»Dnit ; ■ ' ; . ' • 

Thifs target .{Fig. 4) presents groupa of .three characters, two groups 
in a row, with three units of decreir:ent betwean groups, .The. acuity value- 
rar^ge ^'roia £0/40 onallen equivalent to 20/70. The subject vms teated at 
30' and moved forward when neoeauary. 

Two scoring systama vjere effiployed for this target; • 
1» Credit was given for the lest group In which tivo of three char- 

-7- 



acters wers correctly idantified, 

2, Credit was given for the last group prior to the first group in 
which two errors occurred. 

4. CHECKERBOARD ! Thia target, consisting of two plates, is shown in Fig. 
3. The procedure established for its presentation was aa follows: '•In 
the first group of diamonds (squares), upper left, do you see one in which 
there is a checkerboard pattern?" "V;here is it in the second group?" 
"The third groupV'» Yihen 3. is acquainted with the response desired - "All 
right, now tell me where it is in each group. It never will be in the cen- 
ter - always at the top, right, bottom or left. When you are not certain 
as to its location, guess," 

Credit was given for the last group in which 1 error or no error 
was made. Other scoring systems were considered for this test though riot 
employed for reasons mentioned in a later discussion. 

JTATI3TICAL ANALY3I3 

Test-retest reliability coefficients of each test target for the rigjfct 
eye, left eye, and both eyes, respectively, were derived by the produet- 
moment inethod of correlation. Differences between test and retest moans 
were also computed and the statistical significance of the differences was 
determined by the _t teat. 

For the right eye, left eye, and both eyes, respectively, test scores 
on the various targets were intercorrelated. Differences between means on 
the various targets were computed, and the statistical significance of the 
differences was determined by the t test. 

The statistical reliability of all coefficients of correlation was de- 
termined by deriving the ratio between each coefficient and its standard 
error. The mean scores, standard deviation of the distribution, and standard 



error of the mean were obtained for each array of scores. 

In order to avoid difficulties inherent in the correlation and av- 
eraging of ratios, the scores used in all computations were the logaritk- 
mic translations of the Snellen-equivalent scores obtained in the testing 
procedure, 

DI3CU33I0H 
General ; . , 

With the exception of the checkerboard plates and the 3ight Screener, 
there is little practical difference in the overall test-retest reliabil- 
ities of the various targets. The two targets excepted have somewhat low- 
er overall test-retest reliabilities than the remainder. The lowest co- 
efficient for the checkerboard plates, .76, occurs in the tests of the 
left eye, and the lowest coefficient for the Sight Screener, .71, occurs 
in the testa for both eyes. (Table 3) ' 

These, two targets also have the lowest coefficients of correlation 
with the Grow chart, in both monocular and binocular acuity. There is no 
practical difference in the correlation of the remaining targets with the 
Grow teat. (Table 4), 

All retests, monocular and binocular, with all targets result in high- 
er mean scores than were obtained on the tests, and all but three test-re- 
test differences are statistically reliable, CDable 3) The order of these 
differences, however, is so small that little practical significance would 
seem to attach to most of them. The highest difference, which occurs be- 
tween the test and retest means of the checkerboard target for the right 
eye, amounts to b.83 log units, or 2.75 feet in the denominators of the 
Snellen ratios. In other words, not only are test-retest scores on the var- 

-9- 



ioua targets highly correlated, but there is little practical gain in 
acuity scores on the retest. The fact that the largest overall reteat 
gains are shown in the case of the checkerboard plates may be explained 
by the increasing familiarity of the subjects with the target, and the 
consequent discovery by soite of the shift in the selection of cues. 
Whenever this discovery occurred during the retest, a subject was able 
not only to add another line to his credit, but frequently also to read 
all the reisainlng lines of the plate. 

In considering the various targets for actual day-by-day testing, 
It la important to know not only the extent to which they intercorrelate, 
but also the extent to which the acuity scores derived from each are 
comncnsurate with the acuity scores derived from the others. Table 5, 
which presents differences between test means, provides for such com- 
parisons. Table 5 indicates that there is considerable consistency in 
the ranks of the mean scores; i.e., some targets, as scored in this 
study, consistently tend to give higher appraisals of visual acuity than 
others. In general, the highest appraisals of acuity are given by both 
scorings of the Randolph Field 1-unit target and the "easier" first 
scorings of the Randolph Field 3-unit target. The lowest appraisals of 
acuity are given by the checkerboard type plates and the New London tar- 
get, and median appraisals are given by the Grow teat, the Sight Screener, 
and the ''harder'* second scoring of the Randolph Field 3-unit target. 
Most of the differences in mean scores are highly reliable statistically. 
It is doubtful, however, that they have Ircmediate practical importance. 
The largest difference shown in the table, a difference of 14.40 in isiean 
log score between the checkerboard plates and the Randolph Field 1-unit 
plate (first scoring, right eye}, amounts to approxin-Ately 5.5 feet in the 

-10- 



denominators of the LiSnellen ratios. Taking %he Grow teat as a median, and 
compering ite mean scores with tha mean scores of the highest ranking testi 
(iiaridolph Field 1-unit plate, first scoring), the differences reduce to 
apprcxiiifltely 2.5 feet, 3.9 feet, and feet, for right eye, left eye, 
and both eyes, respectively. It wiXi be further noted that it is possible 
to devise different scoring methods which could probably make the order of 
scoi-es of each of the targets approach each other more closely, as was done 
with the two Handolph Field testa in this »tudy. The second scoring of the 
Randolph Field 3-miit target, for example, resulted in mean scores which 
are not significantly different fron the mean scores of the Grow teat. 
Specific : 

Inasmuch as most of the targets ware undergoing an extensive field 
testing for the first time, the design of this project included evalu- 
ation of their ease of administration and objective nature. To facili- 
tate presentation, tests will be discussed individually. 
^' QfiOVJ ! It is evident that tvao factors have been varied in the presen- 
tation of the Grow chart; 

1. A higher illumination (34.0 afc) than prescribed in standard Favy 
procedure was employed, in order to equate the illumination with that of 
the trans-Illuminated targets. This change was adopted merely to obtain 
constant experimental conditions* 

2. Alternate linos were delefced by means of a grid, resulting in a 
wider distance between the test lines. This change was adopted for experi- 
mental convenience, to ellminat© confusion as to line requested and any un- 
due influence of a concentration factor. In the opinion of the examiners, 
the addition of this grid adds appreclafely "td~the ease of administration of 

-11- 



of the Grow chart. 

B. SIGHJ aGRBEEER i 'Phis pi«c© of apparatus presented two difficulties, 
both of which might b« •lixtdnated with a minimuHi of modification. Th« oc- 
clualon by polarization is not coiriplate, allowing a "ghoat" image which re- 
inforces that raeeived by the eye being tested in monocular tests. Sec- 
ondly, there should be some extension on the eye-piece which will hold the 
subject farther froDi the lenses. A regular stereoscope eye-piece with vant« 
for circulation over the face of the lenses would serve to overcome fogging 
from body heat and contact by long eyelashes, 

G, GISJCKERBQARD ; In its present form, this target is apparently neither 
so reliable nor valid as it might be ivith modification. During the testing 
program, it was observed that some subjects may read past the levels at 
which they actually perceive a checkerboard by making a brightness discrim- 
ination. It becan;® evident that this shift in cues or the basis of judgjnent 
was taking place when it was noted that some subjects made frequent errors 
at levels 24/20, 20/20, or 20/17, but read the remainder of the plate per- 
fectly. 

Interrogation of these subjects confirmed the existence of this shift 
in criterion. The change in cues, arid the accuracy of the brightness dis- 
crimination, seem further to ba aided by a configuration fonr,ed by the square 
composing each figure. Two squares which are not the checkerboard square, 
and the middle square form a ''line". This "line" appears to be darker then 
the remainder of the pattern. The subject's choice is then imiriediately res- 
tricted to one of the remaining two squares, and his chances of selectiBg 
the eoiTect one are at least 50/50. The subject then seems to pass through 
a stage in which he decides whether the checkerboard pattern should appear 
lighter or darker than the opposite square. If he decides that the overall 



affect would b« a lighter arta, h» is correct and, though having mad* 
errors at this critical lev«l while passing through thia process,' h« 
will continue throughout the rsKiainder of the test with a perfect or 
noar-parfect scor«. On th« other hand, th® aubject may d«cide that thm 
checkerboard would give a darker area. In thia case, he would be in 
error throughout the rejnainder of the tost. That the above situation 
does obtain became evident vfhen it was noted that the wrong responses 
of the subjects were frequently th© exact opposite of the correct rea- 
ponacs. 

It would seem that these limitations are not basic defects of the 
Ortho-Hater checkerboard technique, but are peculiar to the test plates 
as adapted for trans-illmiiination. Equating the brightness of the 
checkerboard and other squares , possibly by superposing the checker- 
board on a neutral gray background, should reduce the difficulty, 

hMi ^'"^o itenis relative to the I'ew London test are worth 

mentioning. The first is the departure in form of the "K" and the ^^G" 
(Fig. 3) J which contributed an unnecessary amount of confusion to the 
testing situation^ The second item is the nuinber of characters used at 
the higher levels. This target was developed for testing low acuities 
as well as high while the subject remained at 20' distance from the eabl- 
not. However, there are so few characters for testing the low acuities 
that ths raeir.ory factor cannot bo disregarded « Thus, where one or two 
characters are all that are used to test a certain level of acuity, there 
is considerable doubt as to the accuracy at these levels, especially afta 
they have been exposed to one eye. 

^' ^^iMM irlMi: One questionable feature of this test is the 

memonic arrangement of some characters which boeomes doubly important 



when oach character repreaents one unit or level of acuity. The main 
case In point, CTE*, would not b* the aid to the Army personnel on which 
this taat was validated as it was to the Kavy personnel used in this 
study. The reliability and validity shown by this test indicate that 
finer quantitative testing of visual acuity is possible where job onalyaia 
warrants it» 

^' RMDOLHi FIELD 3~0nit: This target is constructed with only 9lx test 
letters, arranged in groups of three. The limited number of letters is 
recognized by some subjects, so they know that the last three letters on 
a line must be some combination of the three not seen in the first group. 
Also, some question might be raised as to the relative facility for mem- 
orizing groups of three characters as opposed to groups of other size, dia- 
counting mnemonic aids. 



* GVE the Navy designation for auxiliary aircraft carriers 

-14- 



1. Adams, J. K. ; Fowler, H. 11,; Imus, H, A, Th« rolatlonahip of visual 
acuity to acuity of otoroaooplc vision, 03RD Report Iv'o. 2087, Applied 
Psychology Panol, WDRC. VVftshington, D, C.s September 15, 1943. 

2, Projsct staff, 03RD Oontract OSIsr-815. SHemoraiidwii No, 12 (to Ap- 
plied Payohology Panel, HDRC)! Intisr-reletlonahipa among seven tests 
of stsreoacopic acuity and the relatlonaliip between two tests of visual 
acuity and two tests of phoriaa. iieahiagton, D, Ct; Karoh 24, 1944. 



TAELS 1 



ACUITIES OF THE TEST POPULATION 
as measured by Grow and 
New London targets 
N a 100 



Right Eye Left i^e Both Eyes 

^ellen 



Sq^uiv, 


Grow 


Hew London 


Grow 


Nevf London 


Grow 


New London 


20/10 


14 




10 


1 


88 


8 


11 




9 




8 




11 


13 




14 




11 




23 


14 




21 




20 




24 


15 


43 




45 




47 




17 




10 




14 




12 


20 


80 


18 


25 


20 


14 


6 




O 




3 




4 




24 


4 




3 




1 




25 




11 




10 




8 


26 


1 


1 


Z 




1 




28 


S 




S 




1 




30 






1 








32 


S 












33 




5 




7 




2 


34 


2 




1 


1 




3 


35 












1 


36 


1 








2 




40 


S 








1 




41 




1 










44 


1 




1 








50 


3 


7 


4 


5 




S 


56 






1 








61 








Z 






67 


2 


1 






1 




83 












1 



TABLE 2 



TARGET ACUITT VALUES 



S|i«llsn 
Equiv. 


T»8t 

1 


Test 

__2 


Tost Test 
3 4 


Tast 
_5 


T«9t 

6 


Log 
Scor«* 


0/200 






X 


X 




00.0 


ICQ 






X 


X 


X 


30.1 


70 








X 




45.6 


50 






X 


X 


X 


60.2 


40 




X 


X 


- . X. , 


X 


69.9 


39 


X 










71.0 


38 


X 








X 


72.1 


37 


X 


X 








73.3 


36 


X 








X 


74.5 


35 


X 










75.7 


34 


X 


X 


X 




X 


76.9 


33 


X 




x(33.3) 






78.3 


32 


X 








X 


79.6 


31 


X 


X 








81. 


30 


X 






X 


X 


82.4 


29 


X 










83.9 


38 


X 


X 


X 




X 


85.4 


27 


X 










87.0 


25 


X 








X 


88.6 




X 


X 


X 






90.3 




X 




X 




X 


92.1 


or? 

23 


X 










93.9 


Sd 


X 


X 






X 


95.9 


PI 


X 










97.9 


20 


X 




X X 


X 


X 


100.0 


19 


X 


X 








102.1 


18 


X 










104.5 


17 


X 




x(16,6)x 






107.2 


16 


X 


X 








109.7 


15 


X 






X 


X 


112.4 


14 


X 




x{14.3)x 






115.5 


13 


X 


X 


x(13,5) 






118.8 


12 


X 










122.3 


11 


X 




... x(iia} 






126.0 


10 


X** 


X 


X X 


X 


X 


130.1 



Teat 1 Randolph Field unit decremejit Test 4 Checkerboard plates 

Test 3 Randolph Pield S-unit " Test 5 Sight Screener 

Test 3 New Lohdon Test 6 Grow-Navy Standard 

* Log score c 100 logj^Q (lO x 1/YA } 
** This test extends to 20/5 



TABLE 3 
TEST - RETE3T MEASURES 









TEST 


RETEST 




M, Diff, 








U 


S,D, 


M 


3.D. 


iHinua T«at v\ r*** 




rt 


103.96 


19.14 


105.14 


18,34 


1.18 


.91 


GROW 


1 


103.97 


17.46 


105.90 


17.14 


2.03** 


.90 




b 


112. si 


14.87 


115,19 


14.49 


2,95** 


,89 


3GREENER 


rt 


103,94 


16.53 


106.92 


14.81 


2.98** 


.84 


1 


105.50 


17,08 


106.80 


15.69 


1.30 


.84 




b 


110.68 


13.27 


113.08 


12.45 


2.40** 


.71 


BOARD 


rt 


98.53 


19.86 


104.36 


20.68 


5.83** 


.86 


1 


100.13 


18.26 


105.03 


19.65 


4.90** 


.76 




b 


108.57 


18.47 


111,75 


17.75 


3.18** 


.81 


LONiX)K 


rt 


101.64 


21.53 


104.71 


20.88 


3,07** 


,92 


1 


102.10 


19,94 


103,36 


20-66 


1.26* 


.93 




b 


111.48 


16.53 


113,68 


15.18 


2.20** 


.92 


RF 

l-»Unit (1) 


rt 


112.93 


19,58 


115.62 


18.72 


2.69** 


,94 


1 


114.48 


16.81 


115,56 


18.23 


1.08 


,92 




b 


121.42 


13.56 


123.18 


12.75 


1.76** 


.90 




rt 


109.29 


22.39 


111.95 


20.10 


2.66** 


,89 


l-Vnit (2) 


1 


110,81 


18.72 


112,70 


20.73 


1,89* 


.87 




b 


119.23 


15.05 


122.02 


13,74 


2.79** 


.85 


RF 

3-Unit (1) 


rt 


105.94 


21.73 


108.88 


20.17 


2.94** 


.93 


1 


106,29 


19.55 


108.70 


19.77 


2.41** 


.88 




b 


114.76 


14.71 


116.60 


14.56 


1,84** 


.85 


RF 


rt 


102,85 


22.82 


106.07 


22,80 


3.22** 


,89 


3-Unit (2) 


1 


108,44 


22.53 


106.74 


21.29 


4.30** 


.86 




b 


112,01 


18.58 


115.19 


15.32 


3.18** 


.83 



(1) First scoring system 

(2) Second scoring aystem 

* Statistically reliable at a £ level greater than .01, but less than ,05. 

** Statistically reliable at a £ level of .01 or less. 

*** All r'3 are statistically reliable at a £ level of ,01 or less. 



TABLE 4 

INTHRCOIffiELATIQNS AMDHG TEST SCORES 



, , , . . Grow 


3.S, 


, Check' bd 


N.L. 










Grow .91 


.76 


.72 


.88 


.89 


.89 


.89 


.86 


Sight Screener 


.84 


.67 


.83 


.77 


.76 


.75 


.78 


Checkerboard 




,86 


.74 


.76 


,73 


,76 


,74 


Neva London 






.92 


.91 


.88 


.90 


.90 


Randolph Field 1-Unit 








.9^4 


.92 


.95 


.90 


Randolph Field 3-Unlt 


-RIGHT 


EYE 






.93 


.91 


,95 


Randolph Field 1-Unit 












.89 


.90 


Randolph Field 3-Unit 


(2} 












.89 




Gtqvi , 90 


.77 


.67 


.89 


.90 


.87 


.87 


.88 


Sight Sereener 


.84 


.48 


.74 


.75 


.77 


.73 


.75 


Checkerboard 




.76 


.67 


.68 


.68 


.67 


,70 


Sew London 






.93 


.90 


.90 


,89 


.88 


Randolph Field 1-Unit 








.92 


.91 


.92 


.87 


Randolph Field 3-Unit 


- LEFT 


EYE 






.88 


.87 


.96 


Handolph Field l-Unit 


(2) 










.87 


.86 


Randolph Field 3-Unit 


(2) 












.86 




Grow ,89 


.76 


.69 


.86 . 


.85 


.82 


.84 


.83 


Sight Sereener 


.71 


.51 


.68 


.72 


.69 


.70 


.73 


Checkerboard 




.81 


.72 


.72 


.70 


.72 . 


.66 


Hew London 






.92 


.87 


.84 


.87 


.83 


Randolph Field 1-Dnit 








.91 


.84 


.92 


.82 


Randolph Field 3-Unit 


— - BOTH EYES 






.85 


.85 


.93 


Rendolph Field 1-Unit 


(£) 








.85 


,83 


Randolph Field S-Unit 


(2) 












. 83 



All r's are statistically reliable at a probability level of less than .01. 



TABLE 5 





Grow 




unecKD' a 


M,L. HJ? 1 Rlr 3 EF 1(3} RF 3(2) 


Mean 


103.96 


103.94 


98.53 


101.64 112,93 105.94 102,29 102,85 


Orow Minus 





0,02 


5,43** 


2.32* -8,97** -1.98* -5,33** 1.11 


Sight If 







5.41** 


2.30* -8.99** -2.00 -5.35** 1.09 


Checker »» 
£afiX£l . 









-3.11*-14.40** -7,41**-10.76** -4.32** 


Naw " 
Loudon 








-11*29** -4.30** -7.65** -1.21 


KF 

l"Unit 




RIGHT 


EYE 


6.99** 3.64** 10,08** 


RF " 

lirVm 






-3.35** 3,09** 










6,44** 




Kean 


103,87 


105.50 


lOo'.lS 102.10 114.48 106,29 110.81 102.44 


C-row Minus 





-1.63 


3,74** 


1,77*-10,61** -2.42** -6.94** 1.43 


Sight " 
Screen or 







-5.37** 


3.40**-8.98** -0.79 -5,31** 3.06* 


Checker n 









-1.97 -14.35** -6.16**-10,6S** -2.31 


New, " 
London 








0. -12.38** -4.19** -8,71** -0.34 


W " 
l~IJnit 




LEFT 


EYE - 


8.19** 3.67** 12.04** 


KF " 






-4.52** 3.85** 


EF " 
1-Cnit 








8.37** 




Mean 


112. S4 


110.68 


108.57 : 


111.48 121.42 114,76 119.23 112,01 


Grrow Einus 





1.56 


3.67** 


0,76 -9.18** -2.52** -6.99** 0.23 


Sight 
Screener 




' 


2.11 


-0.80 -10,74** -4,08** -8.55** -1.33 


Checker >♦ 
Bpard 









-2. 91*- 12. 85** -6.19**-10.66** -3,44* 


N^w It 








^9.94** -3.28** -7.75** -0,53 


SF " 

1-Unit 




- . BOTH 


J3ZE3 ■ 


G 6.66** 2.19** 9.41** 


i^Unit 








-4.47** 2.75** 


RF " 
1-Unit 








7,22** 



Statisticflilly reliable at level greater than ,01 but lesa than .05, 



Statistically reliable at level of .01 or less. 




•H